基于混沌离散PSO算法的乙烯生产过程模型变量的选择

Selection of Modeling Variables Based on Chaotic and Discrete PSO in Process of Ethylene Production

管式炉裂解技术是当今乙烯生产的主要技术,乙烯生产过程中有多种影响质量指标的变量,将全部变量参与建模显然会增加建模难度和复杂性,并且大量的冗余信息会降低模型的精度;针对这一问题,提出了基于混沌思想的离散粒子群(PSO)算法进行模型变量的选择;首先,采用混沌离散PSO算法得到建模的最优输入变量集合,再通过偏最小二乘法(PLS)对所选变量进行建模;结果表明,该方法可更有效地克服传统粒子群算法容易陷入局部最优的问题并建立较高的模型精度。

Tube cracking is the main technology in ethylene production and the quality index is influenced by many variables. Too many variables will obviously result in the increase of complication and difficulty of model structure. Besides, a lot of redundant information will reduce the accuracy of the model. To solve the problem, an algorithm based on chaotic and discrete PSO was proposed in this paper. Optimal input variable combination of ethylene production process is obtained by chaotic and discrete PSO algorithm. Then the model is set up based on the selected variables by PLS algorithm. The results show that the algorithm can make particles avoid falling into local optimum and get a higher accuracy model.